search

awslabs / gap-text2sql

GAP-text2SQL: Learning Contextual Representations for Semantic Parsing with Generation-Augmented Pre-Training
https://arxiv.org/abs/2012.10309
Apache License 2.0
102 stars 25 forks source link

About max_steps #14

Closed Yuan0320 closed 3 years ago

Yuan0320 commented 3 years ago

Hello! Is max_steps set to 41000? The RAT bert version is set to 81000. I want to confirm it. Thank you!

Yuan0320 commented 3 years ago

And after data preprocess, each word of question, table, and column (that is, some words in the vocabulary) has a prefix. Should this exist? Like Ġthe.

JoaoLages commented 3 years ago

I've retrained it and the loss in train is basically 0.00 at step 41000

Yuan0320 commented 3 years ago

I've retrained it and the loss in train is basically 0.00 at step 41000

What's your accuracy in step 41000？

JoaoLages commented 3 years ago

Don't recall it, but I was able to obtain the reported 72.72 EXACT score on dev set

Yuan0320 commented 3 years ago

Don't recall it, but I was able to obtain the reported 72.72 EXACT score on dev set

ok, thanks!